Expansive and Contractive Postures and Movement: A Systematic Review and Meta-Analysis of the Effect of Motor Displays on Affective and Behavioral Responses

Abstract

This review and meta-analysis explores the experimental effects of expansive and contractive motor displays on affective, hormonal, and behavioral responses. Experimental studies were located through systematic literature searches. Studies had to manipulate motor displays to either expansive or contractive displays and investigate the effect of the displays on affect, hormones, or overt behavior. Meta-analyses were conducted to determine the pooled, standardized mean differences between the effects of motor displays on affective, hormonal, and behavioral responses. From 5,819 unique records, 73 relevant studies were identified. Robust differences between expansive and contractive displays emerged for affective responses and overt behavioral responses across contexts, type of manipulation, and methods of measurement. The results suggest that the effects are driven by the absence of contractive motor displays (contractive vs. neutral displays: Hedges’s g = 0.45) rather than the presence of expansive displays (expansive vs. neutral displays: g = 0.06). The findings stand as a corrective to previous research, as they indicate that it is the absence of contractive displays rather than the presence of expansive displays that alters affective and behavioral responding. Future research should include neutral control groups, use different methods to assess hormonal change, and investigate these effects in the context of ideographic goals.

Keywords

methodology quantitative emotion affect expansive displays contractive displays

A massive interest in so-called power posing has sprung up as a result of a cardinal study from 2010 that reported that it was possible to alter not only psychological states and behavior but also hormone levels by assuming different power poses (Carney, Cuddy, & Yap, 2010). The public interest in this research is evident from the more than 50 million views of the corresponding TED talk, asserting that power poses effectively alter behavior and psychological states. However, the findings of this study have been called into question because of the study’s methodological shortcomings and the mixed results from successive replication attempts. The current systematic review and meta-analysis is a synthesis of the available experimental evidence pertaining to the effect of motor displays such as power poses on affective and behavioral responses. This synthesis provides a corrective to previous research, specifically addressing the methodological shortcomings of previous studies and suggesting directions for future research.

The last decade has seen a surge in empirical research exploring the affective and behavioral effects of assuming bodily displays, often believed to reflect an intentional or motivational state, manifested as either engagement with, or withdrawal from, one’s inner or outer world (Eder & Rothermund, 2008; Niedenthal, 2007; Schubert & Koole, 2009; Winkielman, Niedenthal, Wielgosz, Eelen, & Kavanagh, 2014). Such research pertains to the mind–body relationship, which has been heavily debated for centuries (James, 1994). Theories of embodiment and embodied cognition reflect a set of diverse theoretical perspectives, all stressing the importance of the body in human perception and cognition but differing on their inclusion—if at all—of mental representations and cognitive concepts (Damasio, 1999; Shapiro, 2011; Wilson & Golonka, 2013; Wilson, 2002; Winkielman et al., 2014). Specifically, at one end of the spectrum, we find classic cognitive views of information processing that consider the motor system to simply be a peripheral input that shapes the individual’s cognitive input to a mind of abstract symbol processing (e.g., Lakoff & Johnson, 1999). At the other end of the spectrum, we find antirepresentational approaches that stress that the motor system is a perceptual resource in itself (Merleau-Ponty, 1962; Wilson & Golonka, 2013; Wilson, 2002). According to such approaches, the mind must be understood in the context of its relationship to the physical body interacting with the world, in which the body and its perceptually guided motions are believed to do much of the work required to achieve goals, omitting the need for mental representations (Maiese, 2014; Wilson & Golonka, 2013). Across embodiment theories, each granting varying roles to mental representations, the motor system (i.e., postures and movement) has been proposed to affect how we respond to the world affectively and behaviorally (Cesario & McDonald, 2013; Damasio, Everitt, & Bishop, 1996; Michalak, Burg, & Heidenreich, 2012; Niedenthal, 2007; Niedenthal & Brauer, 2012).

Empirical studies on the effect of motor behavior on responses to the world have mainly focused on two overall sets of motor displays. The first set covers happy (e.g., Michalak, Rohde, & Troje, 2015), open (e.g., Latu, Duffy, Pardal, & Alger, 2017), dominant (e.g., Turan, 2015), upright (e.g., Harmon-Jones & Peterson, 2009), and powerful displays (e.g., Garrison, Tang, & Schmeichel, 2016). We term these expansive displays because the body is posed in a way that makes it appear taller and wider than when the body posture is neutral. The second set of motor displays covers withdrawal displays, including submissive (e.g., Welker, Oberleitner, Cain, & Carré, 2013), sad or fearful (e.g., Michalak et al., 2015; Shafir, Taylor, Atkinson, Langenecker, & Zubieta, 2013), closed (e.g., Latu et al., 2017), slouched (e.g., Ceunen, Zaman, Vlaeyen, Dankaerts, & Van Diest, 2014), and powerless displays (e.g., Bailey, LaFrance, & Dovidio, 2017). These displays we term contractive displays because the body is posed in a way that makes it appear shorter and smaller than when the body posture is neutral.

The findings from such studies indicate a reciprocal relationship between the motor system and affective and behavioral outcomes. On the one hand, affect and behavior can modify posture and movement; on the other hand, adopting various motor displays can modify both affective and behavioral outcomes. For example, one study found that walking style (i.e., “happy” vs. “depressed” gait) affected memory bias such that a depressed gait was associated with recalling more negative self-relevant content compared with a happy gait (Michalak et al., 2015). Another study showed that adopting an upright posture caused lower negative affect and anxiety than sitting in a neutral position (Wilkes, Kydd, Sagar, & Broadbent, 2017), and yet another study reported that high-power displays resulted in increased risky gambling behavior (Carney et al., 2010). The findings from these studies suggest that the motor system is not simply a response system that emanates from the individuals’ interaction with their surroundings. Rather, the motor system guides and shapes the perception of the world and therefore is a determining factor in affective and behavioral responding (Damasio et al., 1996; Maiese, 2014; Michalak et al., 2012; Roald, Levin, & Køppe, 2018).

In recent years, expansive and contractive motor displays have received a lot of attention. The effects of these displays were evaluated in the study by Carney and colleagues (2010) that suggested that a certain type of expansive bodily displays, namely power poses, not only result in positive psychological outcomes (e.g., increased feelings of power) but also facilitate a hormonal response associated with approach behavior (e.g., increased testosterone and decreased cortisol) compared with a contractive display (i.e., low-power pose). Since then, studies have replicated the affective effects but despite several attempts have not been able to replicate the hormonal finding. Researchers in the field have therefore found themselves discussing potential moderating factors that could explain this replication failure (e.g., the experimental context, the use of cover stories; Cesario & McDonald, 2013; Garrison et al., 2016; Ranehill et al., 2015). These efforts have led to a body of literature characterized by mixed findings.

A number of reviews and meta-analyses have been conducted in an attempt to synthesize these mixed find-ings:

A narrative review conducted by Carney, Cuddy, and Yap (2015) identified 33 studies of the difference in affective and behavioral responses among expansive motor displays, contractive displays, and neutral displays. The authors concluded that expansive displays are associated with affective changes, including increased power feelings; positive thoughts (Carney et al., 2010); behavioral effects, including increased traffic violations (Yap, Cuddy, Carney, Lucas, & Wazlawek, 2013); and hormonal changes, including increased testosterone and decreased cortisol (Carney et al., 2010).

Simmons and Simonsohn (2017) conducted a p-curve analysis (i.e., a method to detect so-called p-hacking by evaluating the robustness of studies) that suggested an absence of both affective and behavioral effects when taking selective reporting into account.

However, another p-curve analysis conducted by Cuddy, Schultz, and Fosse (2018) using a systematically selected and updated set of studies on power posing led to a different conclusion, namely that expansive postures result in increased power feelings and other positive affective outcomes.

Gronau et al. (2017) investigated the effects of expansive displays across six preregistered studies and found an effect on felt power. On the basis of the results from the meta-analysis on felt power and another preregistered study, it was concluded that, overall, expansive displays do not result in behavioral or hormonal effects but do reliably increase felt power (Cesario, Jonas, & Carney, 2017; Jonas et al., 2017).

Taken together, the syntheses have either not quantified the effect of expansive and contractive movement and posture or have not taken potential moderating factors into account. Although the syntheses have each made important contributions to the literature, a number of crucial methodological differences between studies should be explored as moderators when synthesizing the available evidence. The current study therefore offers a comprehensive, systematic review and meta-analysis of relevant studies comparing the effect of expansive, contractive, and neutral displays on affective and behavioral outcomes, exploring the methodological differences between studies as potential moderators. The following moderators were considered: comparison conditions, type of manipulations of bodily displays, type of outcome assessment, and contextual conditions. The rationale for exploring each of these moderators is presented in the sections that follow.

Comparisons of Bodily Displays

A potential moderator concerns the type of displays that are compared. The cardinal study by Carney and colleagues (2010) compared expansive with contractive motor displays, and several subsequent studies have taken the same approach (e.g., Ranehill et al., 2015; Yap et al., 2013). However, without including a neutral control condition in such experiments, it cannot be determined whether the detected effect is caused by the presence of an expansive display or the absence of a contractive display or vice versa. This point was first mentioned by Credé (2018), who argued that a negative effect of a contractive display should not be taken as evidence for the positive effect of an expansive display. None of the syntheses mentioned above have compared contractive and expansive displays with neutral displays; thus, whether the presence of expansive displays or the absence of contractive displays is responsible for the affective and behavioral effects detected remains undetermined.

Outcome Assessment

Existing syntheses have examined effects across different types of outcomes. These outcomes can be categorized into affective (e.g., power feeling and mood) and behavioral (e.g., risk taking and hormonal change) responses. Affective outcomes are typically evaluated with self-report instruments. This has raised the question of whether the effects of expansive displays on such outcomes reflect a true effect or simply pertain to a methodological artifact caused by the experimental demand of self-report assessment (Cesario et al., 2017). Specifically, the effect may simply reflect an expectation bias from the individuals’ participation in the experiment. Effects detected on more than one type of outcome would therefore indicate more robust findings of the effect of motor displays. Another proposed moderator is therefore the type of outcome assessed. Although previous syntheses have distinguished between these outcomes (Cesario et al., 2017; Jonas et al., 2017), the respective effects have not been quantified.

Context

Cesario et al. (2017) argued that that the question of whether the effect of displays on behavior is to be considered direct or context-dependent is unresolved. Accordingly, another proposed moderator concerns contextual factors. First, there has been a tendency to assume that motor displays directly cause the same affective or behavioral response across contexts and personal goals of the individual (i.e., direct effect; Ekman, 2003; Peper & Lin, 2012). In other words, one would then expect a one-to-one correspondence between the display of the body and a particular affective state or behavior. Such propositions align with an assumption of basic emotions as hardwired programs (Ekman, 1992) that, when activated, manifest themselves as a predetermined set of expressions, including bodily displays, regardless of the context. However, the existence of such programs has been called into question because of a lack of empirical support (Barrett, 2006, 2012). Rather, different affective and behavioral outcomes appear to be able to follow from the same motor displays (Barrett, 2006, 2012). For instance, walking with an expansive posture may cause powerful feelings such as anger or pride but also negative affect such as sadness depending on the particular context (e.g., Cesario & McDonald, 2013). Indeed, studies have found that individuals adopting expansive postures report more negative mood when socially excluded (Welker et al., 2013) and higher depression scores when experiencing failure (Riskind, 1984). Accordingly, it has been proposed that the context may moderate the effects of displays on behavior (Cesario & McDonald, 2013).

Thus, one important contextual factor to consider is the experimental setting. Research points to a number of factors to be taken into account with regard to the experimental setting, including (a) whether the bodily manipulation was conducted in an intra- or interpersonal environment, (b) whether the effect of motor displays was assessed during a task as opposed to immediately after the manipulation, and (c) whether a cover story was used. Concerning the type of context, Cesario and McDonald (2013), for instance, found that expansive postural displays affected only felt power when adopted in an interpersonal context (i.e., when faces were present vs. not present). Concerning the use of an experimental task, Cuddy, Wilmuth, Yap, and Carney (2015) found effects of expansive displays on performance and job hirability. Finally, Carney and colleagues (2015) investigated cover stories as a moderator and suggested that awareness of the hypothesis may influence the result. This suggestion should be further evaluated by means of a meta-analysis to assess the quantitative extent to which this may or may not be the case.

Manipulations of Bodily Displays

Another possible moderator includes how the motor displays are manipulated. Within the realm of experimental studies, manipulations have differed in whether they have instructed participants to adopt certain displays while moving (e.g., Michalak et al., 2015), standing (e.g., Lee & Schnall, 2014), sitting (e.g., Roberts & Arefi-Afshar, 2007), or both standing and sitting (e.g., Smith & Apicella, 2017).

These differences in instructions may have contributed to different effects because some positions may be more efficient than others in facilitating certain outcomes. For instance, it may be more difficult to feel powerful when sitting down in contrast to standing upright. Therefore, whether the effects obtained through different manipulations are comparable is a question that should be investigated. That is, although sitting down may not always result in less powerful feelings, it might make a difference if the effect of sitting down is compared with the effect of standing upright when it comes to feelings of power.

Aims and Objectives

Understanding the involvement of the motor system in affective and behavioral responding is not only important from a basic science perspective. Such knowledge can also be translated into a clinical setting, potentially offering a supplement to existing treatments in changing dysfunctional affective and behavioral responding and fostering the emotional changes needed for an individual to take adaptive action. This meta-analysis therefore aims to review the affective and behavioral effects detected in studies on expansive and contractive displays and to explore whether these effects differ between the proposed moderating conditions.

Method

Inclusion and exclusion criteria

Eligible studies were required to be randomized, experimental studies investigating the effect of whole-body motor displays (i.e., posture or movement) on affective outcomes (e.g., mood, self-efficacy, affective content of memory retrieval) or behavioral outcomes (e.g., risk taking, behavioral hormones such as testosterone). The included studies had to examine at least two groups that had been randomly assigned to different experimental motor-manipulation conditions (e.g., different postures or motor manipulations vs. neutral control group). In addition, studies had to include a sample of healthy adults (at least 18 years old) and use a manipulation of the motor system with the aim of inducing an expansive or contractive posture or type of movement.

Studies were excluded if (a) the manipulations of motor displays could not be meaningfully categorized as expansive or contractive (e.g., simply standing, sitting, or lying); (b) only physiological outcome measures (e.g., blood pressure and breathing rate) were obtained, because such outcomes are affected directly by the position of the body (e.g., blood pressure is different when standing vs. lying); (c) only isolated body parts were manipulated (e.g., head or arm), which could not be characterized as whole-body movement, so manipulations could not be considered manipulations of posture or movement (this exclusion criterion of isolated body-part manipulations is consistent with previous syntheses of the literature; see Cuddy et al., 2018); (d) if motor manipulations were combined with nonmotor manipulations (e.g., expansive posture and positive thoughts vs. contractive posture and negative thoughts), because such manipulations would render a determination of the effect of the motor manipulations impossible; and (e) the outcomes were not affective or behavioral (e.g., simple math performance or neutral word retrieval).

Studies including additional experimental conditions (e.g., manipulation of negative or positive mood or of social inclusion or exclusion in addition to manipulation of the motor system) were not included in the meta-analyses because it would be impossible to isolate the effects of the manipulation of the motor system from the effects of the additional manipulations. However, to ensure the breadth and comprehensiveness of the current study, a narrative synthesis was conducted for these studies to tease out conditional effects (i.e., interaction effects indicating that the effect of manipulation of the motor system on behavioral outcomes depends on an additional manipulation).

Literature searches

A keyword-based search in the electronic databases of PsycINFO and PubMed was conducted to locate studies from the earliest dates available until May 2019. In the final search string, keywords related to the motor system (gait OR posture OR postural OR embodiment OR embody OR “gross motor system” OR “motor system” OR “body movement” OR “movement pattern*” OR “submissive posture” OR “subordinate posture” OR “contractive posture” OR “slumped posture” OR “slumped position” OR “slouch* posture” OR “slouch* position” OR “stooped posture” OR “stopped position” OR “slow gait” OR “slumped gait” OR “upright gait” OR “dominant posture” OR “upright posture” OR “upright position” OR “expansive posture”) were combined with keywords related to affect and behavior (”negative emotion*” OR “positive emotion*” OR “positive mood” OR “negative mood” OR “negative affect” OR “positive affect” OR depressive* OR “anxiety*” OR “psychological distress” OR distress* OR mood OR emotion* OR power* OR “action tendenc*” OR “emotion* reactivity” OR “emotion* arousal” OR confidence OR dominance OR power OR attitude OR “memory bias” OR “self-efficacy” OR “risk taking” OR self-esteem OR “emotion* stress” OR motivation* OR preference* OR “behave* hormone*” OR hormon* OR testosterone OR cortisol). In order to limit records falling within our inclusion criteria, we actively excluded a number key words: Parkinson* OR Alzheimer* OR dementia OR stroke OR syndrome OR knee OR surgery OR “case report” OR lesion OR chronic OR cancer OR drug* OR injuri* OR sclerosis OR autism OR “hip fracture” OR “heart attack” OR concussion OR “qualitative study” OR mice OR rats OR “back pain” OR creutzfeldt OR transcranial OR “lumbar spinal stenosis” OR “cystic fibrosis” OR diabetes OR cerebral palsy OR sarcopenia OR hypotension OR amputati* OR patient OR disorder. The searches were limited to English-language, peer-reviewed records. Reverse searches both in reference lists of located studies and in previous syntheses of the literature (Carney et al., 2015; Simmons & Simonsohn, 2017) were conducted. In addition, a forward search in more recent publications citing the located studies were conducted until no additional relevant articles were found.

Data extraction and coding procedures

Two independent reviewers assessed the eligibility of articles through title and abstract screening. The reviewers then screened full texts of eligible articles to assess whether they should be included. In the case of disagreements, a third party (i.e., the last author) was consulted. The first and last author extracted data from all eligible studies together. Studies were coded for type of motor display (i.e., movement or posture), type of control group (i.e., neutral or contractive), type of experimental context (e.g., neutral, success feedback, or interpersonal), cover story (yes or no), experimental effect (immediate or during a task), and type of task if relevant (e.g., gambling or performance). Studies were also coded for type of outcome, including self-report, overt behaviors, or behavioral hormones. Self-reported outcomes included affective outcomes that were further coded according to valence (i.e., positive self-reported outcomes and negative self-reported outcomes). In addition, studies were coded for body position (e.g., sitting vs. standing), participant characteristics (i.e., mean age and gender distribution), and time spent in the manipulated motor condition.

Effect sizes (ESs) were expressed as Hedges’s g, which is a variant of Cohen’s d that accounts for small sample sizes (Cohen, 1988; Hedges, 1981). When possible, ESs were computed using means and standard deviations. Alternatively, ESs were calculated on the basis of other test statistics (e.g., N and p). If no relevant test statistics were available, the study authors were contacted. When authors were unresponsive, nonsignificant results were included with an ES of 0 to retain all relevant studies. In total, eight of 235 effects were coded as 0.

Analytic overview

Meta-analyses were performed to determine the pooled, overall effect of manipulating the motor display (e.g., posture, gait) on the dependent variable (e.g., mood, behavioral hormones). Main analyses included three separate meta-analyses comparing expansive and contractive displays, contractive and neutral displays, and expansive and neutral displays and were based on random-effects models, which allow for heterogeneity in the analysis of the overall difference between groups (Borenstein, Hedges, Higgins, & Rothstein, 2010). Effects were averaged within and across outcomes such that all studies, relevant for the particular analysis, were represented with only one ES to meet the assumption of independence between observations (Borenstein et al., 2010).

In addition, when an adequate number of studies (k ≥ 3) was identified, planned exploratory moderation analyses were performed. Moderation analyses of categorical variables (e.g., type of manipulation, outcome, and task) were explored with meta-analysis of variances (ANOVAs). Because of the small number of studies in some of these analyses, we compared the differences in effects between the largest and smallest ESs obtained for the particular moderating categories. Moderation analyses of continuous variables (i.e., age, percentage women, time in motor display) were performed with metaregression analyses, estimated with the maximum-likelihood method. In these regression analyses, the proposed moderators operated as independent variables predicting the ES. All moderation analyses were based on random-effects models.

Heterogeneity (i.e., the degree to which ESs vary across studies) was explored using Q and I² statistics. Because of frequent reports of low statistical power for heterogeneity tests, a more liberal p value (≤ 0.10) was used to establish significant heterogeneity (i.e., Q; Poole & Greenland, 1999). The I² value indicates how much of the variation is due to difference in true effects (Higgins, Thompson, Deeks, & Altman, 2002).

To evaluate the risk and impact of publication bias, funnel plots (i.e., a graphic illustration of study ESs in relation to study size) were visually expected, and an Egger’s test was performed to statistically evaluate the distribution of the ESs (Egger, Smith, Schneider, & Minder, 1997). If the results of the Egger’s test indicated funnel-plot asymmetry (e.g., by pointing to a lack of small studies with small ESs), an adjusted ES was calculated using the trim-and-fill method (Duval & Tweedie, 2000). ESs of missing studies were estimated and imputed using this method, resulting in a new adjusted ES. To evaluate the robustness of the results, fail-safe numbers were calculated for statistically significant results (Deeks, Macaskill, & Irwig, 2005; Rosenthal, 1991). Rosenthal’s fail-safe number indicates the number of unpublished studies with null findings that would reduce the combined effect to a nonstatistically significant level. A fail-safe number above 5 × k + 10, where k refers to the number of studies included in the meta-analysis, is considered sufficiently robust (Rosenthal, 1991). We conducted adjustments only when there was a risk of publication bias (i.e., when the fail-safe N was not met or when the Egger’s test was significant). All analyses were performed with Comprehensive Meta-Analysis, Version 3.3.070 (http://www.meta-analysis.com).

Results

Search results and characteristics of studies

The study-selection process is depicted in Figure 1. The initial literature search produced 8,000 records with 2,181 duplicates. Of these, 5,584 records were excluded on the basis of abstract screening, and 235 underwent a full-text screening. During this screening, the most common reason for exclusion was a lack of relevant motor manipulations (n = 60), followed by manipulations of facial expressions alone or in combination with posture (n = 51) and the sole use of emotional manipulations (n = 15), respectively. After reviewing full-text articles, 52 records with 73 independent studies were included for systematic review. The final sample of studies included a total of 7,038 participants and were published in scientific journals between 1982 and 2019. Of the 73 studies, 48 were subjected to meta-analysis. The remaining 25 studies randomized participants to additional conditions (e.g., success vs. failure experience in expansive vs. contractive postures) and were therefore subjected to a narrative synthesis.

Fig. 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart of the study-selection process.

Characteristics of studies included in the meta-analysis

Study characteristics are provided in Table 1. The majority of studies manipulated body posture (k = 40), followed by movement (k = 7) and body balance (k = 1). Most studies compared expansive and contractive displays (k = 40), followed by contractive and neutral displays (k = 8) and expansive and neutral displays (k = 6). The most commonly used outcome was self-report measures (k = 39), followed by overt behavioral measures (e.g., gambling choice; k = 29). Twenty studies used both methods (k = 20). The least-frequent outcome was behavioral hormones (k = 4). Most studies investigated motor manipulations in neutral contexts (k = 29), followed by contexts with faces present (k = 9) and contexts characterized by success (k = 3), failure/criticism (k = 3), dyadic interactions (k = 2), and expectancy (k = 2). Twenty studies assessed the effect of bodily manipulations immediately after the motor manipulation (e.g., mood). In 34 studies, participants were exposed to a task, the most frequent type being gambling tasks (e.g., money gambling; k = 9), followed by choice/moral-dilemma tasks (e.g., revenge; k = 7), performance tasks (e.g., job interview; k = 5), judgment or social judgment tasks (e.g., liking of others; k = 5), cognitive tasks (e.g., math problem solving; k = 4), memory tasks (e.g., affective memory bias; k = 2), and administrative tasks (i.e., restricted eating; k = 1). Seven studies evaluated the effect both immediately after the manipulation and after engaging in a task.

Table 1.

Overview of Included Studies

Study	Total (N)	Women (N)	Mean age (years)	Manipulation type and duration^a	Immediate effect^b	Task performance (behavior)^c	Setting^d	Context^e	Cover story	Additional manipulation^f
Allen et al. (2013)	97	97	19.61	Post: C vs. E standing vs. N sitting T: 30 min	—	Administrative (restricted eating: 2)	Medical monitor attached	Neutral □	Marketing and physiology	—
Arnette & Pettijohn (2012)	42	30	22.43	Post: C (negative) vs. E (positive) sitting T: 1 min	—	Choice task (preference: 2)	—	Neutral □	Marketing research	—
Bailey et al. (2017)	94	49	20.6	Post: C (low-power) vs. E (high-power) sitting and standing T: 2.5 min	—	Gambling (risk-taking behavior: 2; power feelings: 1.2)	—	Faces present ■	Effect of posture on cognitive performance	—
Bialobrzeska & Parzuchowski (2016) – Study 2	70	60	21.79	Post: C vs. 2 × N (simple vs. complex) standing T: 0.5 min	—	Social-judgment & compliance task (perceived social dominance: 1.2; behavioral submissiveness: 2)	—	Neutral ■	Testing a new laser device for measuring body	—
Bialobrzeska & Parzuchowski (2016) – Study 3	39	22	22.22	Post: C vs. N (simple) T: 0.5 min	Submissiveness (norm compliance: 1.2; spatial distance: 2; gratification delay: 2; mood: 1.2; self-esteem: 1.2)	—	Fake body measures	Neutral ■	Testing a new laser device for measuring body	—
Bohns & Wiltermuth (2012) – Study 1	89	44	—	Post: C (submissive sitting) vs. E (dominant standing) vs. N (standing)	Pain threshold: 2; perceived dominance: 1.2	—	Blood-pressure measures (cuff)	Neutral □	Health benefits of exercise at work	—
Bombari et al. (2017)	200	101	20.83	Post: C vs. E standing T: 2 min	—	Gambling task (risk taking: 2; power feelings: 1.2)	—	Faces present ■	Separate studies (body posture/first impressions & gambling)	—
Briñol et al. (2009)	71	—	—	Post: C (doubtful) vs. E (confident) sitting T: ~1.5 min	Traits relative to future performance (attitudes: 1; thoughts: 1; confidence: 1; mood: 1; task difficulty: 1)		—	Negative or positive thoughts □	Separate studies (job satisfaction & performance)	Thought direction (positive vs. negative)
Carney et al. (2010)	42	26	—	Post: C (low-power) vs. E (high-power) sitting and standingT: 1 min (2 min total)	—	Gambling task (risk taking: 2; power feelings: 1.2; hormones: 2)	ECG leads	Faces present ■	Physiological study	—
Cesario & Johnson (2018) – Study 1	164	122	—	Post: E (sitting and standing) vs. N (no pose)T: 1 min (2 min total)	Power feelings: 1.2	Gambling task (risk taking: 2; negotiation: 2)	Watching a TED talk	Expecta-tions ■	No (effect of holding positions on power and negotiation)	—
Cesario & Johnson (2018) – Study 2	161	0	—	Post: E (sitting and standing) vs. N (no pose)T: 1 min (2 min total)	Power feelings: 1.2	Gambling task (risk taking: 2; negotiation: 2)	Watching a TED talk	Expecta-tions ■	No (effect of holding positions on power and negotiation)	—
Cesario & Johnson (2018) – Study 3	71	—	—	Post: C vs. E sitting and standingT: 1 min (2 min total)	Power feelings: 1.2	Gambling task (risk taking: 2; negotiation: 2)	—	Dyadic interactions ■	Blind to purpose	—
Cesario & Johnson (2018) – Study 4	174	—	—	Post: C vs. E sitting and standingT: 1 min (2 min total)	Power feelings: 1.2	Gambling task (risk taking: 2)	—	Dyadic interactions ■	Blind to purpose	—
Cesario & McDonald (2013) – Study 1	209	—	—	Post: C vs. E standing and sitting vs. N (no pose)T: 1 min (2 min total)	—	Gambling task (risk taking: 2)	—	Faces present or not ■	Physical body and memory vs. forming impressions	No faces vs. faces present
Cesario & McDonald (2013) – Study 2	155	—	—	Post: C vs. E standing and sittingT: 1 min (2 min total)	—	Gambling task (risk taking: 2)	—	Role ■	Physical body and role	Imaginary role: submissive vs. dominant
Ceunen et al. (2014)	36	29	19.44	Post: C (slouched) vs. E (upright) vs. N sittingT: 4 min each	Mood: 1.1, 1.2	—	EMG electrodes attached	Neutral □	—	—
Cuddy et al. (2015)	61	40	—	Post: C (low-power) vs. E (high-power) standingT: 5–6 min	—	Performance task (hirability: 2; performance: 2; power feelings: 1.2)	—	Neutral ■	Physical motion and performance	—
de Zavala et al. (2017)	82	64	22.99	Post: C (low-power standing and sitting) vs. E (high-power standing) vs. open yoga pose (standing) vs. closed yoga pose (standing)T: 2 min	Subjective energy: 1.2; self-esteem: 1.2	—	Mock electrodes attached	Faces present ■	Fatigue and social perception	—
Fischer et al. (2011) – Study 2	36	22	20.83	Post: C (closed) vs. E (open) sitting	Power feelings: 1.2	Choice task (investment in organic foods & decision: 2)	—	Neutral □	—	—
Flack (2006)	52	32	—	Post: C (afraid & sad) vs. E (happy & angry) sittingT: 10 s	Mood: 1.1, 1.2	—	Electrodes attached	Neutral □	purpose explicitly told	—
Förster & Stepper (2000) – Study 1	40	—	—	Post: E (open) standing vs. C (closed) kneeling downT: ~3 min	—	Memory task (correctly recalled words: 2; mood: 1; subjective valence of words: 1)	Tango music	Affective □	Ergonomic research	Positive vs. negative words
Garrison et al. (2016)	305	187	18.82	Post: C (low-power) vs. E (high-power) sitting and standingT: 1 min each (2 min total)	—	Gambling task (risk taking: 2; power feelings: 1)	Sensors attached	Gaze direction □	Posture’s effect on mental processes and physiology	Gaze direction (direct vs. averted)
Hackford et al. (2019)	73	—	—	Move: C (slumped) vs. E (upright) walking postureT: 13 min	—	Performance task (mood: 1.1, 1.2; pain: 1.1; sleepiness: 1.1; power feelings: 1.2)	Physiology measured	Neutral ■	Effect of walking on cognition	—
Hao et al. (2017) – Study 1	149	60	20.34	Post: C (closed) vs. E (open) standingT: 3 min each (6 min total)	—	Creativity task (emotional arousal: 1; valence: 1; power feelings: 1)	—	Affective □	—	Mood induction (emotional videos)
Hao et al. (2017) – Study 2	130	85	22.06	Post: C (closed) vs. E (open) standingT: 3 min each (6 min total)	—	Creativity task (emotional arousal: 1; valence: 1; power feelings: 1; persistence: 2)	—	Affective □	—	Mood induction (emotional videos)
Harmon-Jones & Peterson (2009)	46	23	—	Post: C (reclined) vs. E (upright)T: ~2 min	Mood: 1	—	Essay on smoking in public	Affective ■	Personality’s effect on essay content and impression of others	Insult vs. no insult
Huang et al. (2011) – Study 1	77	53	—	Post: C vs. E sittingT: 3–5 min		Mood-related task (power words chosen: 2; power feelings: 1)	Puzzle task	Role ■	Marketing test on ergonomic chairs	Role (manager vs. subordinate)
Huang et al. (2011) – Study 2	77	43	—	Post: C vs. E sittingT: 3–5 min		Gambling task (action taking: 2; power feelings: 1)	Abstraction & puzzle task	Role ■	Decision making & gestalt completion	Role (manager vs. subordinate)
Huang et al. (2011) – Study 3	57	30	—	Post: C vs. E sittingT: 3–5 min	—	Moral dilemma (action taking: 2)	—	Role ■	Marketing on ergonomic chairs & memory task	Recalled power (high vs. low)
Jackson et al. (2017) – Study 1	65	65	—	Post: C vs. E standing and sittingT: 1 min each (2 min total)	Self-concept: 1.2	—	Physiological monitors attached	Faces present ■	Physiological study	—
Jackson et al. (2017) – Study 2	128	128	20.75	Post: C vs. E sitting and standingT: 1 min each (2 min total)	Self-concept: 1.2; psychological needs: 1.2; social status: 1.2; willingness to communicate: 1.2; mood: 1.1, 1.2; psychological flexibility: 1.2; emotion regulation: 1.2; perceived power: 1.2; vitality: 1.2; internalizing: 1.1; rumination: 1.1; perspective: 1.1; self-objectification: 1.1	—	Physiological monitors attached	Faces present ■	Physiological study	—
Keller et al. (2017)	292	191	—	Post: C (low-power) vs. E (high-power) sitting and standingT: 1 min each (2 min total)	—	Gambling task (risk taking): 2; performance task (hirability): 2; performance: 2; power feelings: 1)	—	Faces present ■	Only for unaware (physiological study)	Aware (watching TED talk) vs. unaware
Kille et al. (2012)	47	22	21.08	Balance: C (unstable) vs. E (stable) sitting	—	Social-judgment task (judgment of instability: 1.1; desire for stability: 1.1; mood: 1.1, 1.2)	—	Neutral □	—	—
Klaschinski et al. (2017)	200	133	24.46	Post: C (low-power) vs. E (high-power) standingT: 6 min	—	Performance task (hirability: 2; speech performance: 2; social sensitivity: 2; power feelings: 1.2)	Saliva samples	Neutral ■	Physical motion and performance	—
Kozak et al. (2014)	80	80	20	Post: C (slumped) vs. E (upright) sittingT: 5 min	—	Cognitive task (mood: 1; personal statements: 1; traits: 1)	Cognitive task	Success feedback ■	Effect of ergonomics (posture) on physiological responses	Chair (throne vs. child’s) & shirt (tank top vs. sweatshirt)
Latu et al. (2017)	200	111	22.43	Post: C (closed) vs. E (open) standing	—	Judgment task (attitude: 1; power feelings: 1; thought confidence: 1; openness: 1)	Neutralization procedure (attitude)	Faces present ■	Evaluation of messages on global issues	Persuasion (weak vs. strong arguments)
Lee & Schnall (2014) – Study 2	41	17	28.24	Post: C (low-power) vs. E (high-power) sittingT: 3 min	—	Judgment task (weight judgment: 2)	—	Neutral □	Ergonomics of work environments	—
Michalak et al. (2015)	39	22	21.27	Move: C (depressed) vs. E (happy)T: 8 min	—	Memory task (affective memory bias: 1.1)	Biofeedback	Neutral □	Gait patterns in response to real-time feedback	—
Mussweiler (2006) – Study 2	37	—	—	Move: C (biking at slow speed) vs. N (normal speed)T: 5 min	—	Social judgment task (ratings of person: 2)	—	Neutral ■	Unrelated studies	—
Mussweiler (2006) – Study 3	33	—	—	Move: C (walking at slow speed) vs. N (normal speed)T: 5 min	—	Social judgment task (ratings of words: 2)	—	Neutral ■	Unrelated studies	—
Nair et al. (2015)	74	41	26	Post: C (slumped) vs. E (upright) sittingT: 30 min	Mood: 1.1, 1.2	Performance task (perceived threat: 1.1; valenced language use: 2; self-esteem: 1.2; task persistence: 2)	Finger cuff	Neutral ■	Effect of tape on upper back on physiology	—
Naruse & Hirai (2000) – Study 2	10	10	20.3	Move: C (slow tempo) vs. N (preferred tempo)T: 0.5 min (6 min total)	Anxiety: 1.1	—	—	Neutral □	—	—
Park et al. (2013) – Study 2ag	213	111	—	Post: C vs. E standingT: 3 min	Power feelings: 1.2	—	—	Faces present ■	Postures for pilot study	—
Park et al. (2013) – Study 2bg	119	78	—	Post: C vs. E sittingT: 3 min	Power feelings: 1.2	—	—	Neutral □	Ergonomic chair quality testing	—
Park et al. (2013) – Study 3g	106	42	—	Post: C (sitting) vs. E (feet-on-desk-posture)T: 3 min	Valenced language use: 2; power feelings: 1.2	—	—	Neutral □	Ergonomic chair quality testing	—
Park et al. (2013) – Study 4g	83	37	—	Post: C (sitting) vs. E (feet-on-desk-posture)T: 3 min	—	Moral dilemma (risk taking: 2)	—	Neutral □	—	—
Peper & Lin (2012)	100	73	23.7	Move: C (slumped) vs. E (engaged) gaitT: 2–3 min	Subjective energy: 1.2	—	—	Neutral □	—	—
Peper et al. (2017)	216	142	24.6	Post: C (slumped) vs. E (upright) sittingT: 1.5 min each (3 min total)	—	Memory task (affective recall ease: 1.1, 1.2)		Neutral □	—	—
Ranehill et al. (2015)	200	98	—	Post: C (low-power) vs. E (high-power) sitting and standingT: 3 min each (6 min total)	—	Gambling task (risk taking: 2; hormones: 2; power feelings: 1.2)	—	Neutral □	No: physical position effect on hormone levels and behavior	—
Riskind (1984) – Study 1	76	38	—	Post: C (slumped) vs. E (upright) vs. N (no instructions) sittingT: 8 min	—	Cognitive task (locus of control: 1; expectancy of success: 1)	Puzzles	Success or failure feedback ■	Separate spatial thinking study & biofeedback study	Success vs. failure feedback
Riskind (1984) – Study 2	51	0	—	Post: C (slumped) vs. E (upright) vs. N (uncomfortable/distracting) sittingT: 8 min	—	Cognitive task (persistence: 2;depression: 1; locus of control: 1)	Spatial thinking test	Success or failure feedback ■	Separate spatial thinking & biofeedback studies	Success vs. failure feedback
Riskind (1984) – Study 3	20	0	—	Post: C (slumped) vs. E (upright) sittingT: 8 min	—	Cognitive task (depression: 1.1; locus of control: 1.2)	Insolvable puzzles	Failure feedback ■	Separate spatial thinking & biofeedback studies	—
Riskind & Gotay (1982) – Study 1	20	0	—	Post: C (slumped) vs. E (upright) sittingT: 8 min	—	Cognitive task (persistence: 2; self-confidence: 1.2; feelings of strength: 1.2; weakness: 1.1; tiredness: 1.1)	Spatial thinking test	Success feedback ■	Separate spatial thinking & biofeedback studies	—
Riskind & Gotay (1982) – Study 2	20	11	—	Post: C (slumped) vs. E (upright) sittingT: 8 min	—	Cognitive task (persistence: 2; self-confidence: 1.2; feelings of strength: 1.2; weakness: 1.1; tiredness: 1.1)	Spatial thinking test	Success feedback ■	Separate spatial thinking study & biofeedback study	—
Riskind & Gotay (1982) – Study 4	41	0	—	Post: C (tensed) vs. N (relaxed) sitting	Mood: 1; emotional stress: 1; expected attributions of performance: 1	Cognitive task (persistence: 2)	Spatial thinking test	Affective ■	Effect of posture on nerve-impulse firing	High vs. low threat (awaiting IQ test)
Roberts & Arefi-Afshar (2007)	60	47	—	Post: C (slumped) vs. E (upright) sittingT: 3 min	—	Cognitive task (mood: 1.1, 1.2)	Cognitive task	Success feedback ■	Separate cognitive & physiological studies	—
Ronay et al. (2017)	108	64	21.48	Post: C (low-power) vs. E (high-power) standing and sittingT: 1 min (2 min total)	—	Gambling task (power feelings: 1.2; risk taking: 2; hormones: 2; overconfidence: 1.2)	Sensors attached	Faces present ■	Physiological study	—
Rossberg-Gempton & Poole (1993)	24	12	—	Post: 3 × C (closed) vs. 3 × E (open) sitting and standingT: 15–30 s each (3 min total)	Mood: 1	—	—	Expectancy □	—	4× expectancy conditions (no, neutral, directive, valenced)
Shafir et al. (2013)	22	11	25.4	Move: C (sad & fearful) vs. E (happy) vs. NT: 0.5 min	Mood: 1.1, 1.2	—	Movement practicing & motor-imagery task	Neutral □	Specific muscles and brain-function interest	—
Smith & Apicella (2017)	247	0	22.0	Post: C (low-power) vs. E (high-power) vs. N sitting and standingT: 63 s (2 min 6 s total)	—	Physical task (testosterone: 2; cortisol: 2; risk taking: 2; power feelings: 1)	—	Faces present ■	Role of biology in decision making	Winner vs. loser
Stepper & Strack (1993) – Study 1	99	0	—	Post: C (slumped) vs. E (upright) vs. N sitting		Cognitive task (mood: 1; pride: 1)	Cognitive & evaluation task	Success feedback ■	Influence of different ergonomic positions on different tasks	Posture onset (three time points)
Strelan et al. (2014) – Study 3	85	50	—	Post: C (powerless) sitting vs. E (powerful) standingT: 2 min	—	Moral dilemma (revenge: 1.1)	—	Neutral □	Separate studies on bodily experiences	—
Tiedens & Fragale (2003) – Study 2	80	46	20	Post: C vs. E sitting	—	Social judgment task (liking of confederate: 1)	Sensors attached	Position ■	Sharing of unique information & physiological information	Other people’s posture (expansive vs. constrictive)
Turan (2015)	85	0	—	Post: C (submissive) vs. E (dominant) sittingT: 22 min	—	Performance task (hormone: 2; interpersonal dominance: 1.2)	Electrodes attached	Failure feedback ■	Cardiac activity	—
Veenstra et al. (2017) – Study 1	229	163	21.05	Post: C (slumped) vs. E (upright) vs. N (comfortable) sitting T: 4 min	—	Mood regulation task (mood generation: 1; recovery: 1; valenced thoughts: 1)	Imagination task	Neutral or negative mood ■	Effect of body posture on performing a task	Neutral vs. negative-mood induction
Veenstra et al. (2017) – Study 2	122	75	20.45	Post: C (slumped) vs. E (upright)	—	Memory task (mood generation: 1; recovery: 1; memory bias: 2)	Imagination task	Negative mood ■	Effect of body posture on performing a task	Emotion regulation (reappraisal vs. spontaneous)
Welker et al. (2013) – Study 1	91	56	—	Post: C (submissive) vs. E (dominant) sitting	—	Virtual game (threat on basic needs: 1; mood: 1)	—	Role ■	No	Role: included vs. excluded
Welker et al. (2013) – Study 2	84	51	—	Post: C (submissive) vs. E (dominant) sitting	—	Virtual game (threat on basic needs: 1; mood: 1)	—	Role ■	No	Role: included vs. excluded
Wilkes et al. (2017)	61	43	26.26	Post: E (upright) vs. N sittingT: 25 min	Mood: 1.1, 1.2	Performance task (mood: 1.1, 1.2; valenced language use: 2)	—	Neutral ■	Effect of physiotherapy tape on cognitive task	—
Wilson & Peper (2004)	24	13	32	Post: C (slumped) vs. E (upright) sittingT: 1 min (4 min total)	Generation ease (valenced thoughts: 1.2)		Sensors attached	Neutral □	—	—
Yap et al. (2013) – Study 1	88	31	—	Post: C vs. E standingT: 1 min	—	Moral dilemma (stealing: 2)	Survey (impression of best friend)	Faces present ■	Effects of stretching on impression formation	—
Yap et al. (2013) – Study 2	34	20	—	Post: C vs. E sittingT: 7 min	—	Moral dilemma (cheating: 2)	Solved anagrams	Rude feedback □	Effect of feng shui on creativity	—
Yap et al. (2013) – Study 3	71	48	—	Post: C vs. E sittingT: 5 min	—	Moral dilemma (traffic violations: reckless behavior & hit and runs: 2)	—	Neutral □	Physiology and video games	—

Manipulation refers to posture (Post) or movement (Move). Abbreviations used in this column: C = contractive; E = expansive; N = neutral; T = time spent in motor display. If time is not included, it was not available.

Immediate effect assessment occurred after the manipulation, coded as follows: 1 = self-report; 1.1 = negative self-reported outcomes; 1.2 = positive self-reported outcomes; 2 = objective.

Task performance refers to the effect assessed during/after a task, coded as follows: 1 = self-report; 1.1 = negative self-reported outcomes; 1.2 = positive self-reported outcomes; 2 = objective.

Setting refers to the setting before manipulation. Abbreviations used in this column: EMG = electromyography.

Context refers to the context during manipulation. Words used in this column: Affective = mood induction; Dyadic interactions = interaction with confederate; Expectancy = expectancy of manipulation effect; Faces present = faces present during the manipulation; Neutral = no manipulation before/during motor manipulation; Position = position relative to confederate; Role = assignment to an imaginary or real role; ■ = interpersonal; □ = intrapersonal.

Words used in this column: Success/rude/failure feedback = feedback from experimenter before motor manipulation.

In these studies, effect sizes were calculated across individuals from different cultures.

Meta-analyses

Overall pooled ESs and between-studies differences

The pooled effects from the overall random-effects meta-analyses are provided in Table 2. Three overall, combined main effects across contexts and outcome measures were obtained. Both the expansive and contractive (k = 40, Hedges’s g = 0.40; p = < .001; 95% confidence interval [CI] = [0.29, 0.50]) and contractive and neutral (k = 8, Hedges’s g = 0.45, p < .001, 95% CI = [0.23, 0.67]) displays showed significant differences in the overall combined effect, and the fail-safe numbers indicated robustness of these results. However, the overall combined effect for expansive versus neutral displays was nonsignificant (k = 6; Hedges’s g = 0.06; p = .197; 95% CI = [−0.09, 0.20]). Because of the small number of studies comparing contractive and neutral displays (k = 8), moderation analyses were performed on only the 40 studies comparing expansive and contractive displays (see Fig. 2).

Table 2.

Pooled Effect Sizes Across Affective and Behavioral Outcomes and Moderator Variables

Outcome	Sample size (k)	Heterogeneity				Global effect sizes			Fail-safe N	Criterion
Outcome	Sample size (k)	Q	df	p	I ²	Hedges’s g	95% CI	p	Fail-safe N	Criterion
Overall combined effect
Expansive vs. contractive	40	256.1	39	< .001	84.8	0.40	[0.29, 0.50]	< .001	2,741	210
Adjusted for publication bias	40					0.40	[0.29, 0.50]
Expansive vs. neutral	6	15.6	5	.008	68.0	0.06	[−0.09, 0.20]	.197
Contractive vs. neutral	8	12.4	7	.088	43.5	0.45	[0.23, 0.67]	< .001	53	40
Type of context
Neutral	23	167.2	22	< .001	86.8	0.45	[0.29, 0.62]	< .001	1,117	125
Faces present	9	34.2	8	< .001	76.6	0.29	[0.13, 0.46]	< .001	94	55
Adjusted for publication bias	11					0.19	[0.02, 0.36]
Success feedback	3	4.9	2	.087	59.1	0.36	[0.05, 0.67]	.025	10	25
Adjusted for publication bias	3					0.36	[0.05, 0.67]
Critical feedback	3	2.3	2	.313	14.0	0.27	[−0.04, 0.58]	.083
Between-groups difference: all		1.29	3	.731
Between-groups difference: neutral vs. critical feedback		1.0	1	.306
Type of outcome
Self-report	33	173.1	32	< .001	81.5	0.45	[0.32, 0.57]	< .001	1868	175
Adjusted for publication bias	33					0.45	[0.32, 0.57]
Negative outcomes	14	61.0	12	< .001	78.7	0.37	[0.14, 0.59]	.001	162	80
Positive outcomes	31	134.5	30	< .001	77.7	0.45	[0.32, 0.58]	< .001	1328	165
Adjusted for publication bias	35					0.36	[0.21, 0.50]
Behavioral	22	116.6	21	< .001	82.0	0.36	[0.18, 0.54]	< .001	382	120
Hormone	4	8.4	3	.039	64.2	0.14	[−0.13, 0.41]	.307
Testosterone	3	5.4	2	.068	62.8	−0.02	[−0.41, 0.38]	.942
Cortisol	4	4.2	3	.237	29.1	0.23	[−0.01, 0.47]	.065
Immediate or task effect
Immediate	17	90.5	16	< .001	82.3	0.47	[0.29, 0.66]	< .001	465	95
Adjusted for publication bias	19					0.41	[0.23, 0.58]
Task	27	176.6	26	< .001	85.3	0.38	[0.24, 0.51]	< .001	1152	145
Cover story
Yes	31	161.2	30	< .001	81.4	0.37	[0.27, 0.48]	< .001	1573	165
No or —	9	93.8	8	< .001	91.5	0.48	[0.13, 0.83]	.007	152	55
Between-groups difference		0.3	1	.557
Type of manipulation
Posture	35	245.4	34	< .001	86.1	0.38	[0.26, 0.49]	< .001	1981	185
Movement	4	3.9	3	.275	22.6	0.52	[0.32, 0.72]	< .001	36	30
Between-groups difference		1.5	1	.219
Position of motor display
Sitting	18	81.4	17	< .001	79.1	0.46	[0.28, 0.64]	< .001	673	100
Standing	7	32.4	6	< .001	81.5	0.34	[0.11, 0.57]	.004	57	45
Sitting and standing	11	55.0	10	< .001	81.8	0.29	[0.13, 0.44]	< .001	139	65
Movement	4	3.9	3	.275	22.6	0.52	[0.32, 0.72]	< .001	36	30
Between-groups difference: sitting and standing vs. moving		3.4	1	.065
Type of task
Gambling	7	34.5	6	< .001	82.6	0.27	[0.07, 0.46]	.007	55	40
Performance	5	50.8	4	< .001	92.12	0.38	[0.08, 0.67]	.012	110	35
Moral dilemma	5	7.9	4	.094	49.6	0.47	[0.18, 0.76]	.002	22	35
Adjusted for publication bias	5					0.47	[0.18, 0.76]
Cognitive task	4	6.0	3	.114	49.6	0.33	[0.03, 0.62]	.029	9	30
Adjusted for publication bias	5					0.38	[0.12, 0.63]
Between-groups difference: all		1.2	3	.760
Between-groups difference: moral dilemma vs gambling		1.2	1	.264

Note: faces present = faces present during the manipulation; immediate effect = effect assessed immediately after the manipulation; neutral = no manipulation before/during motor manipulation; Success/critical feedback = feedback from experimenter before/during motor manipulation; type of context = context during posture manipulation; task effect = effect assessed during/after a task.

Fig. 2.

Results from a meta-analysis depicting differences between contractive and expansive displays. The analysis was performed with Comprehensive Meta-Analysis software (Version 3.3.070; www.meta-analysis.com).

Moderator outcome assessment

Regarding the type of outcome (i.e., self-report, behavior, or hormones) and averaging across other potential moderators, the results indicated significant and robust overall effects for all self-reported outcomes combined (k = 33, g = 0.45, p < .001; adjusted for publication bias), negative self-reported outcomes (k = 14, g = 0.37, p = .001), and positive self-reported outcomes (k = 35, g = 0.36, p < .001; adjusted for publication bias). In addition, the results indicated significant and robust effects for overt behavioral outcomes (k = 22, g = 0.36, p < .001). This was not the case for behavioral hormones (k = 4, g = 0.14, p = .307). Given that most studies used more than one type of outcome, it was not possible to analytically compare effects among the different types of outcome, as this would violate the assumption of observation independence. The obtained effects thus have to be interpreted according to their numerical size. To investigate these results further, and because much attention has been paid to the effect of expansive postures on hormones (e.g., Credé & Phillips, 2017; Ranehill et al., 2015; Ronay, Tybur, van Huijstee, & Morssinkhof, 2017), we evaluated this effect separately for the two different types of hormones assessed, namely testosterone and cortisol. Although there was a trend toward a significant ES for cortisol (k = 4, g = 0.23, p = .065), such an effect was not found for testosterone (k = 3, g = −0.02, p = .942), suggesting that cortisol, but not testosterone, may be influenced by changes in bodily displays. These effects should, however, be viewed in light of the low number of studies investigating behavioral hormones.

Context

Studies were set in varying types of contexts, defined as the overall setting for the experiment to which all participants were exposed. Averaging across other potential moderators, results indicated significant effects within most context types, including a neutral context (k = 23, g = 0.45, p < .001), an interpersonal context (i.e., faces present) (k = 9, g = 0.29, p < .001; adjusted for publication bias: k = 11, g = 0.19), and a success context (k = 3, g = 0.36, p = .025; adjusted for publication bias). However, no significant effect was evident for the context in which participants received critical feedback, where only a trend toward significant difference was obtained (k = 3, g = 0.27, p = .083), which should be interpreted in light of the low number of studies in this condition. The between-groups difference between studies was nonsignificant, suggesting that context did not moderate the overall effect significantly. However, because the critical-feedback context yielded a nonsignificant effect, we explored whether the between-groups difference was statistically different by comparing this context with the neutral context, which obtained the largest effect. This between-groups difference was not significant (Q = 1.29, p = .731).

A second contextual moderator was then explored, namely whether the outcome was obtained either immediately after the motor manipulations or in response to a task. Statistically significant and robust ESs were obtained both for immediate assessments (k = 17, g = 0.47, p < .001; adjusted for publication bias: k = 19, g = 0.41) and delayed assessments (i.e., after a task; k = 27, g = 0.38, p < .001) as evident from the fail-safe numbers exceeding the criteria. The effects were comparable in size, indicating no difference in ESs depending on when the results were obtained.

Concerning the type of task (e.g., gambling, performance), statistically significant and robust effects were found across types—gambling tasks (k = 7, g = 0.27, p = .007), performance tasks (k = 5, g = 0.38, p = .012), moral-dilemma tasks (k = 5, g = 0.47, p = .002; adjusted for publication bias), and cognitive tasks (k = 4, g = 0.33, p = .029, adjusted for publication bias: k = 5, g = 0.38). There was no statistical significant difference between the condition with the smallest ES (i.e., gambling) and that with the largest ES (i.e., moral dilemma; p = .264).

The last contextual moderator was whether studies incorporated a cover story. The meta-analysis suggested robust effects both for studies with cover stories (k = 31, g = 0.37, p < .001), and studies without cover stories (k = 9, g = 0.48, p = .007). There was no statistically significant difference in these effects (Q = 0.3, p = .557), which suggests that the overall effect did not differ according to whether studies included a cover story.

Manipulations of bodily displays

The last planned moderation analyses considered manipulations of bodily displays, including the type of manipulation. Averaging across other potential moderators, both posture, (k = 35, g = 0.38, p < .001) and movement (k = 4, g = 0.52, p < .001) yielded statistically significant and robust effects; however, there was not a significant difference between the two groups. In addition, the position of the body suggested robust effects for all studies, a finding that was supported by the fail-safe number, including sitting (k = 18, g = 0.46, p < .001), standing (k = 7, g = 0.34, p = .004), sitting and standing combined (k = 11, g = 0.29, p < .001), and movement (k = 4, g = 0.52, p < .001). A between-groups difference between the largest and smallest ESs (for standing and moving, respectively), however, was not statistically significant.

Continuous moderators

For three additional moderators (i.e., mean age, gender distribution, and time spent in the motor displays), metaregression-based moderation analyses indicated no significant moderating effects of age (k = 18, b < 0.06, SE = .03, p = .057), gender (k = 38, b = −0.01, SE = .01, p = .122), or time spent in motor display (k = 36, b < −0.01, SE = .01, p = .821).

Narrative review of studies with additional manipulations

To further explore the effect of the motor system on behavioral outcomes, we narratively reviewed the studies that used motor manipulations in combination with other manipulations (e.g., negative vs. positive mood or failure vs. success experience). In these studies, a significant interaction effect between the manipulation of motor displays and the additional manipulation suggest that the impact of motor displays on behavioral outcomes was conditional on the additional manipulation. We reviewed the interaction effects narratively only if three or more studies used the same type of additional manipulation, that is, explored the same conditional effects.

Of all studies, which included an additional manipulation, 45.8% reported significant interaction effects. The studies could be categorized into three categories: social role, affective conditions, and success and failure experience.

Social role

Eight studies investigated the conditional effect of posture (i.e., expansive vs. contractive) depending on role assignment (e.g., submissive vs. dominant or excluded vs. included). Two studies reported significant interactions, two studies reported mixed results, and four studies found no such interaction effects.

Regarding the positive interaction effects, one study investigated whether the effect of participants’ posture (expansive vs. contractive) was conditional on a confederate’s posture (expansive vs. contractive) in relation to liking the confederate (p < .05). The results suggested that participants in the expansive condition liked the contractive confederate more than the expansive confederate (the participants rated on a 7-point scale how much they liked the confederate and how popular they assessed the confederate to be). The opposite pattern held true for the participants in the contractive condition, in which they liked the expansive confederate more than the contractive confederate (Tiedens & Fragale, 2003, Study 2). The other study explored the interaction between posture (expansive vs. contractive vs. neutral) and human interaction (faces present vs. faces not present) on risky gambling and found a significant effect (p = .048), suggesting that the effect of posture was conditional on the presence of faces. Specifically, with faces present, participants in the expansive posture gambled more than participants in the contractive condition, but this was not the case when faces were not present (Cesario & McDonald, 2013, Study 1).

Concerning the studies with mixed effects, two studies investigated the interaction between posture (dominant vs. submissive) and role (inclusion vs. exclusion) and reported significant effects for mood (p = .021, Study 1; p = .026, Study 2). These interactions indicated that the effect of posture on mood was conditional on role. Specifically, the findings showed that (a) excluded participants who assumed dominant postures reported higher levels of negative mood than excluded individuals who assumed submissive postures, and (b) excluded participants reported higher levels of negative mood than included participants in dominant-posture conditions (Welker et al., 2013). However, the same two studies reported diverse findings in relation to the interaction effect between posture and role on felt threats to basic needs. Whereas Study 1 showed no significant interaction (p = .265), Study 2 reported a significant interaction effect (p = .003), suggesting that excluded participants assuming a dominant posture had lower threats to basic needs compared with participants assuming a submissive posture (Welker et al., 2013).

Regarding the nonsignificant interaction effects, four studies considered the interaction between posture (expansive vs. contractive) and hierarchical role (submissive vs. dominant) on power feelings, use of power-related words, action taken (Huang, Galinsky, Gruenfeld, & Guillory, 2011, Studies 1–3), and risky gambling (Cesario & McDonald, 2013, Study 2), respectively. None of the six possible interaction effects reached statistical significance, suggesting that the effect of posture on behavior was not conditional on role in these studies (Cesario & McDonald, 2013; Huang et al., 2011).

Affective conditions

Seven studies investigated whether the effect of posture was conditional on affective manipulations (positive vs. negative conditions), and only two studies reported significant interactions.

Concerning the studies reporting a significant interaction effect, one study showed a significant interaction effect between posture (defined as confident vs. doubtful) and thought content (positive vs. negative) on attitudes (p < .01). Specifically, this study showed that confident postures combined with negative thoughts resulted in less encouraging attitudes than doubtful postures combined with negative thoughts. In contrast, confident postures combined with positive thoughts resulted in more encouraging attitudes than doubtful postures combined with positive thoughts (Briñol, Petty, & Wagner, 2009). Another study showed a significant interaction effect between posture (constrictive vs. neutral) and threat manipulation (high vs. low threat) on resulting perceived task difficulty (p = .03). The results suggested that contractive postures combined with high levels of threat and neutral postures combined with low levels of threat resulted in more perceived task difficulty, whereas contractive postures combined with low levels of threat and neutral postures combined with high levels of threat resulted in less perceived task difficulty (Riskind & Gotay, 1982, Study 4). However, the same study showed no interaction effect between posture and thought direction on persistence with insolvable puzzles (Riskind & Gotay, 1982).

Five other studies showed no significant interaction effects. Two of these studies investigated the interaction between posture (open vs. closed) and emotion (positive vs. negative) on resulting power feelings (Hao, Xue, Yuan, Wang, & Runco, 2017, Studies 1 and 2). Another study considered the interaction between posture (standing upright vs. kneeling down) and word valence (positive vs. negative) on mood and subjective perceived valence of words (Förster & Stepper, 2000, Study 1). A fourth study investigated the interaction between posture (upright vs. reclined) and insult (insult vs. no insult) on mood (Harmon-Jones & Peterson, 2009), and the fifth study explored the interaction between posture (straight vs. stooped) and mood (negative vs. neutral) on the valence of thoughts (Veenstra, Schneider, & Koole, 2017, Study 1).

Success and failure experience

Three studies investigated the interaction effect between assuming different postures (e.g., expansive vs. contractive) and success or failure experiences. All three studies found significant interaction effects on some, but not all, outcome variables.

Specifically, one study reported significant interaction effects between posture (upright vs. slumped) and experience (success vs. failure) on locus of control (p < .03) and expectancy of success (p < .05; Riskind, 1984, Study 1). The results indicated that an upright posture combined with a failure experience caused participants to experience more external locus of control than when in a slumped posture. In contrast, a slumped posture caused participants in the success condition to experience more external locus of control than when in an upright posture. In addition, expectations of success were higher when the upright posture was combined with success experience and lower when upright posture was combined with failure experience.

Another study showed significant interactions between posture (upright vs. slumped vs. neutral) and experience (success vs. failure) on persistence (p < .01) and depression (p < .05) but not locus of control (Riskind, 1984, Study 2). In this study, failure experiences combined with upright postures and success experiences combined with slumped postures caused higher levels of depression compared with failure experiences combined with slumped postures and success experiences combined with upright postures. In addition, failure experiences combined with slumped postures and success experiences combined with upright postures caused greater persistence than failure experiences combined with upright postures and success experiences combined with slumped postures (Riskind, 1984).

The last of the three studies found no significant interactions between posture (high power vs. low power vs. neutral) and experience (success vs. failure) on power feelings (p = .634), economic risk taking (p = .850), and cortisol (p = .216; Smith & Apicella, 2017). However, the study reported a significant interaction between competition and posture on testosterone (p = .011), demonstrating that expansive posture combined with success manipulations and contractive posture combined with failure manipulations increased testosterone, whereas expansive posture combined with failure manipulations and contractive posture combined with success manipulations decreased testosterone (Smith & Apicella, 2017).

Additional conditions

The remaining seven studies investigated the interaction between posture (i.e., expansive vs. contractive) under the following conditions: (a) onset of posture manipulation (four time points; Stepper & Strack, 1993, Study 1), (b) gaze direction (i.e., direct or averted; Garrison et al., 2016), (c) persuasion (i.e., strong or weak messages from the researchers; Latu et al., 2017), (d) knowledge of proposed benefits of power poses (i.e., aware vs. unaware; Keller, Johnson, & Harder, 2017), (e) expectancy of effect (i.e., no, neutral, directive, valenced; Rossberg-Gempton & Poole, 1993), and (f) emotion regulation (i.e., reappraisal vs. spontaneous regulation; Veenstra et al., 2017, Study 2). Moreover, one study added two additional manipulations (i.e., size of chair and size of shirt; Kozak, Roberts, & Patterson, 2014).

Discussion

The findings of this meta-analysis stand as a corrective to previous conclusions that the experimental effects of expansive displays (e.g., power poses) on affective and behavioral responses are beneficial. First, the results indicate that the experimental effect of expansive and contractive displays on affective and behavioral outcomes largely seems to be caused by the absence of contractive displays rather than the presence of expansive displays. Specifically, the absence of contractive displays was associated with more positive outcomes, whereas there appeared to be no effect of the presence of expansive displays. This was evident from the statistically significant main effects identified for contractive versus neutral displays (g = 0.45) and expansive versus contractive displays (g = 0.40) but not for expansive versus neutral displays (g = 0.06). The latter finding is especially noteworthy, as most studies have previously assumed that expansive displays are responsible for the detection of differences between the effect of expansive and contractive displays on affective and behavioral responses (e.g., Carney et al., 2010; Cuddy et al., 2015). The results of the current meta-analysis challenge this assumption and emphasize the importance of the criticism from Credé (2018), specifically that a negative effect of contractive displays should not be taken as evidence for a positive effect of expansive displays.

Together, these results can be taken as preliminary evidence of the impact of contractive displays on affective and behavioral responses. Although we identified a substantial difference in ES (i.e., small to moderate) between expansive and contractive displays compared with a neutral control group, these analyses were based on the small number of studies (n = 14) that included a neutral control group. Caution must therefore be taken when interpreting the findings because of the small sample size. This underscores the importance of replicating this finding and exploring it further in future research.

A number of moderators have previously been proposed that this meta-analysis evaluated statistically.

Outcome assessment

There was a statistically significant difference between expansive and contractive displays in overall effects on self-reported outcomes (g = 0.45), such as power feelings (Cesario et al., 2017; Gronau et al., 2017). This was also the case when positive self-reported outcomes (g = 0.36) and negative self-reported outcomes (g = 0.37) were analyzed separately. As opposed to previous suggestions (Jonas et al., 2017), the results from this meta-analysis also suggest a statistically significant and robust difference in effects of motor displays on overt behavioral outcomes (g = 0.36). One explanation for this finding may be that the current meta-analysis included more studies than the seven preregistered studies found in Jonas et al. (2017). With the value of preregistering studies (i.e., preventing scientific exploration in the case of null results) in mind, future authors of preregistered studies should determine whether such behavioral effects can be replicated. Replicating previous findings (Credé & Phillips, 2017; Jonas et al., 2017), the results of the meta-analyses conducted separately for different types of outcomes indicated no difference between motor displays regarding hormonal effects (g = 0.14). Because of the considerable interest in behavioral hormones (Carney et al., 2010; Credé & Phillips, 2017), we compared the effect of motor displays on cortisol and testosterone separately. Given the current finding that the absence of contractive displays rather than the presence of expansive displays may drive affective and behavioral effects, it is notable that the effect on cortisol was trending toward significance (k = 4, g = 0.23, p = .065). Cortisol has frequently been associated with emotional distress and avoidance behavior (Ma, Abelson, Okada, Taylor, & Liberzon, 2017; Roelofs et al., 2009; Young & Korszun, 2010), possibly speaking to the importance of the absence of hormones associated with distress and avoidance rather than the presence of hormones previously associated with approach tendencies, such as testosterone (Frye & Rhodes, 2008; Josephs, Sellers, Newman, & Mehta, 2006). However, this should be viewed in light of the small number of studies and small ES.

The finding that the effects seem to be restricted to constrictive displays deserves specific attention, as it counters previous assumptions that the effect should be attributed to the presence of expansive displays. One particular point stressed within embodiment theories is the importance of analyzing the task at hand, asking how the body may or may not be an important resource in the particular situation, thus pointing to a need to map the involved operations and specific context (Wilson & Golonka, 2013). When conducting a thorough analysis of the task at hand and specific context, researchers should be explicit about their theoretical perspective concerning embodied cognition.

Contextual factors

Regarding the previously investigated contextual factors, the results of the meta-analysis suggest statistically significant effects within all contextual conditions, including types of contexts, tasks, and cover stories (g = 0.27–0.47). Only small numerical differences between the various moderating conditions for context were identified. Hence, the current findings are at first glance incongruent with the notion that motor displays exert an effect only in specific contexts, that is, a context-dependent effect (e.g., only in tasks or only with faces present but not in neutral conditions; Cesario & McDonald, 2013). However, it could indeed be argued that the experimental demand, which all participants were exposed to, pertains to the context. In this view, there is no such thing as a “neutral context,” implying that the experimental demands have likely played a nontrivial role in the included studies. In addition, if participants perceived the experimental setting as safe and unthreatening, this may have affected the experimental outcome.

Considering such experimental demands in the interpretation, the results of this meta-analysis would counter an interpretation of the effect detected as a universal, context-independent effect of the contractive displays. Indeed, such an effect seems unlikely when further taking the narrative review into account. Specifically, the narrative review emphasized the importance of context-congruent tasks. Five studies indicated that incongruent conditions (i.e., conditions not matched on seemingly congruent features, e.g., the adoption of expansive postures when experiencing personal failure or exclusion) caused maladaptive outcomes. For instance, one study found that incongruent conditions caused significantly higher levels of depression and lower persistence in puzzle solving than congruent conditions (i.e., failure condition/contractive posture and success condition/upright posture; Riskind, 1984). Another study demonstrated that congruent conditions (i.e., expansive posture combined with success manipulations and contractive posture combined with failure manipulations) increased testosterone, whereas incongruent conditions decreased testosterone (Smith & Apicella, 2017), thus demonstrating the congruency effect not only at the affective and overt behavioral level but also, biologically, at the hormonal level. This finding is in line with one synthesis of three studies that reported evidence for a so-called mismatch effect regarding experienced status (i.e., dominant vs. submissive) and testosterone level (high vs. low; Josephs et al., 2006). Results showed that the experience of having a dominant status combined with low testosterone level resulted in worse cognitive performance than with the experience of having a submissive status combined with low testosterone. At the same time, when a submissive status was combined with a low testosterone level, better cognitive performance was obtained than when it was combined with a high testosterone level (Josephs et al., 2006).

Combined, such findings question the inherent positive or negative effect of expansive and contractive postures or movement, respectively. Rather, these findings align with the idea of context-dependent effects. Thus, although we detected an effect of motor displays on affective and behavioral outcomes across contexts in the meta-analysis, the narrative review found that the effects depended on contextual congruency. These combined findings point to the importance of a thorough analysis of both the task and surrounding context at hand. In conducting this analysis, the following considerations should be taken into account.

First, individual differences may cause the same experimental context to be perceived differently (e.g., a highly perfectionistic person may report more negative outcomes in performance tasks than others may). Thus, although using a randomized procedure when assigning participants to the different experimental conditions, individual differences may result in varying perceived within-condition contexts, the mean of which is not necessarily meaningful considering the detected importance of context congruency. Future studies should therefore consider individual differences that are likely to affect the perception of the task and context. In addition, it is likely that not all experimental goals are perceived to be similarly relevant across participants because individuals may have divergent goals and current concerns (Scherer & Moors, 2019). All studies in this review indeed relied on experimentally manipulated tasks or imposed goals (e.g., simulated job-performance tasks), and one may question how relevant and important the participants perceived the experimentally manipulated tasks or goals to be. For example, would performance at a job interview be perceived as important for the participant with no interest in the job or with low levels of achievement motivation? If this is not the case, it would be difficult to specifically test the idea that motor displays are dependent on the context if what actually matters is how the context is perceived by the individual, that is, personal relevance (Wilson & Golonka, 2013). One way of addressing this issue is by designing experiments in which the task concerns the individual’s idiographic goals, thereby exploring the role of motor displays in relation to personal attainments.

Manipulations of bodily displays

Regarding different manipulations of bodily displays, one moderation analysis explored the type of manipulation (i.e., posture vs. moving). The results suggested robust effects for both types of manipulations; however, there was a numerical difference in effects in favor of moving (g = 0.52) compared with posture (g = 0.38). The moderation analysis exploring motor-display position led to similar results in that the group difference between sitting/standing and moving was almost significant in favor of moving (p = .065). However, this result was only trending toward significance and should thus be further explored to determine whether moving in certain ways results in greater effects than changing posture. Once again, we would like to underscore the importance of conducting a task analysis that considers when standing versus sitting, or moving versus assuming a certain posture, may or may not be an important resource in the particular situation.

Research implications

On the basis of the results of this comprehensive meta-analysis, we advise that future researchers take into account the current findings and the exposed shortcomings of past studies. Doing this, we believe, will truly advance the field beyond simple replications of the findings of past research. We advise future researchers to consider the following three points. First, the results point to the importance of including a neutral control condition in experimental studies of the effect of manipulating motor displays, allowing the effect to be ascribed to the appropriate condition. By including a neutral control group, future researchers will be able to further investigate the effects of contractive displays on affective and behavioral responses.

Second, because of the small number of studies on the impact of motor displays on hormonal effects, future research should further investigate this matter. In doing so, researchers may benefit from using more optimal measures of hormonal changes, including blood samples as opposed to saliva samples as well as more assessment points in contrast to only pre- and postassessments.

Third, future studies should take demand characteristics and contextual factors into account. This may be achieved by including a suspicion check of manipulations of bodily displays or extending manipulations beyond experimentally induced goals to address individuals’ idiographic goals, thereby exploring the role of motor displays in relation to personal attainments. No previous studies have evaluated participants’ personal goals. However, doing this could shed light on the real-world relevance of assuming different motor displays. In addition to this, investigating effects of posture or movement patterns in naturalistic conditions would be an important next step.

Clinical implications

The findings are clinically relevant in a number of ways. Should these findings, supporting the importance of absence of contractive displays over the presence of expansive displays, be replicated, they can begin to inform clinical practice. Indeed, preliminary studies have shown that motor manipulation can affect dysfunctional affective processes in depressed patients (Michalak, Chatinyan, Chourib, & Teismann, 2018; Michalak, Mischnat, & Teismann, 2014). If this finding turns out to be robust, we can, for instance, start thinking about how to cultivate the practice of assuming motor displays as an integral part of psychotherapy. Specifically, the change of contractive motor displays (i.e., from slumped to neutral or upright) may be a promising technique with which to assist the regulation of dysfunctional affective states and behaviors. In psychotherapy for emotional disorders (i.e., anxiety and mood disorders), this knowledge is especially crucial because current therapies, although having improved their effectiveness, still leave a large minority of clients with suboptimal treatment responses (Farabaugh et al., 2012; Hofmann, Asnaani, Vonk, Sawyer, & Fang, 2012; Newman et al., 2011). Although existing cognitive treatment programs involve techniques with which to direct clients’ attention to their body (e.g., for breathing techniques from mindfulness-based therapies, see Kabat-Zinn, 2005; for relaxation techniques and exposure exercises from cognitive behavioral therapy, see Beck, 2013), these interventions do not explicitly address the idea that motor displays may constrain affective and behavioral responses and thus may be used as an explicit change strategy.

In addition, the findings suggest that psychotherapies may benefit from addressing congruency between displays, bodily state, and mental state. On the basis of the narrative review, it seems important to avoid incongruency (e.g., expansive posture combined with failure experiences; c.f. Riskind, 1984; Smith & Apicella, 2017; Welker et al., 2013). In this congruency view, it would not be beneficial to adopt a simple so-called power pose (c.f. Cuddy et al., 2015) when feeling upset. Acknowledging the importance of congruency and idiographic goals, it seems that the implementations of motor displays as an intervention tool should be performed cautiously.

Limitations

First, one limitation pertains to between-studies heterogeneity. Although effects were evaluated within different categories of moderators, the between-studies heterogeneity for most categories was large. This indicates that potentially unidentified variables were responsible for systematic variation in the results. Second, the results may have limited real-world relevance. The current meta-analysis included only experimental studies that explored the role of motor displays within a laboratory setting. Although such a methodology has the advantage of controlling for potential confounding variables, it remains uncertain whether results from experimental settings are generalizable to real life. Therefore, it is necessary to explore the effects of motor displays outside the laboratory. Third, although we included only studies that manipulated “whole-body” movement, there is an arbitrary line between posture and, for example, simple head movement, which may also affect posture. Future studies could parse out effects by investigating how much isolated body movement is needed to create an effect. Fourth, five authors did not respond to our requests for missing data, possibly introducing bias into the results. Fifth, the aim of the current meta-analysis was to determine the average ES for group differences in the effect of motor displays on affective and behavioral responses and to investigate potential moderators influencing the effect. As moderation analyses were conducted only for studies comparing expansive and contractive displays, we lack studies investigating these matters, including a neutral control group. That is, different moderating conditions may possibly exist for expansive and contractive displays, respectively.

Conclusion

The current meta-analysis denotes a systematic attempt to synthesize the existing evidence regarding the effect of motor displays on affective and behavioral responses. The results condense a heterogenic body of experimental research and suggest that the absence of contractive displays is a stronger driver of the effects detected than the presence of expansive displays. However, this suggestion should be viewed in light of the small sample of studies comparing expansive and neutral conditions as well as contractive and neutral conditions. The findings indicate overall robust differences between expansive and contractive displays across contexts, types of manipulations, and methods of measurement, except for behavioral hormones. Overall, the findings of the current meta-analysis draw attention to various avenues for clinical work and future research.

Footnotes

Transparency

Action Editor: Laura A. King

Editor: Laura A. King

ORCID iDs

Emma Elkjær

Mai B. Mikkelsen

References

*Allen

Gervais

S. J.

Smith

J. L.

(2013). Sit big to eat big: The interaction of body posture and body concern on restrained eating. Psychology of Women Quarterly, 37, 325–336. doi:10.1177/0361684313476477

*Arnette

S. L.

Pettijohn

T. F.

(2012). The effects of posture on self-perceived leadership. International Journal of Business and Social Science, 3, 8–13. Retrieved from http://ijbssnet.com/journals/Vol_3_No_14_Special_Issue_July_2012/2.pdf

*Bailey

A. H.

LaFrance

Dovidio

J. F.

(2017). Could a woman be superman? Gender and the embodiment of power postures. Comprehensive Results in Social Psychology, 2, 6–27. doi:10.1080/23743603.2016.1248079

Barrett

L. F.

(2006). Are emotions natural kinds? Perspectives on Psychological Science, 1, 28–58.

Barrett

L. F.

(2012). Emotions are real. Emotion, 12, 413–429. doi:10.1037/a0027555

Beck

J. S.

(2013). Kognitiv adfærdsterapi [Cognitive behavioral therapy] (pp. 23–66). København, Denmark: Akademisk Forlag.

*Bialobrzeska

Parzuchowski

(2016). Size or openness: Expansive but closed body posture increases submissive behavior. Polish Psychological Bulletin, 47, 186–194. doi:10.1515/ppb-2016-0022

*Bohns

V. K.

Wiltermuth

S. S.

(2012). It hurts when I do this (or you do that): Posture and pain tolerance. Journal of Experimental Social Psychology, 48, 341–345. doi:10.1016/j.jesp.2011.05.022

*Bombari

Schmid Mast

Pulfrey

(2017). Real and imagined power poses: Is the physical experience necessary after all? Comprehensive Results in Social Psychology, 2, 44–54. doi:10.1080/23743603.2017.1341183

10.

Borenstein

Hedges

L. V.

Higgins

J. P. T.

Rothstein

H. R.

(2010). A basic introduction to fixed-effect and random-effects models for meta-analysis. Research Synthesis Methods, 1, 97–111. doi:10.1002/jrsm.12

11.

*Briñol

Petty

R. E.

Wagner

(2009). Body posture effects on self-evaluation: A self-validation approach. European Journal of Social Psychology, 39, 1053–1064. doi:10.1002/ejsp.607

12.

*Carney

D. R.

Cuddy

A. J. C.

Yap

A. J.

(2010). Power posing: Brief nonverbal displays affect neuroendocrine levels and risk tolerance. Psychological Science, 21, 1363–1368. doi:10.1177/0956797610383437

13.

Carney

D. R.

Cuddy

A. J. C.

Yap

A. J.

(2015). Review and summary of research on the embodied effects of expansive (vs. contractive) nonverbal displays. Psychological Science, 26, 657–663. doi:10.1177/0956797614566855

14.

*Cesario

Johnson

D. J.

(2018). Power poseur: Bodily expansiveness does not matter in dyadic interactions. Social Psychological and Personality Science, 9, 781–789. doi:10.1177/1948550617725153

15.

Cesario

Jonas

K. J.

Carney

D. R.

(2017). CRSP special issue on power poses: What was the point and what did we learn? Comprehensive Results in Social Psychology, 2, 1–5. doi:10.1080/23743603.2017.1309876

16.

*Cesario

McDonald

M. M.

(2013). Bodies in context: Power poses as a computation of action possibility. Social Cognition, 31, 260–274. doi:10.1521/soco.2013.31.2.260

17.

*Ceunen

Zaman

Vlaeyen

J. W. S.

Dankaerts

Van Diest

(2014). Effect of seated trunk posture on eye blink startle and subjective experience: Comparing flexion, neutral upright posture, and extension of spine. PLOS ONE, 9(2), Article e88482. doi:10.1371/journal.pone.0088482

18.

Cohen

(1988). Statistical power for the behavioral sciences. Hillsdale, NJ: Erlbaum.

19.

Credé

(2018). A negative effect of a contractive pose is not evidence for the positive effect of an expansive pose: Commentary on Cuddy, Schultz, and Fosse (2018) [Abstract]. Retrieved from https://ssrn.com/abstract=3198470

20.

Credé

Phillips

L. A.

(2017). Revisiting the power pose effect: How robust are the results reported by Carney, Cuddy, and Yap (2010) to data analytic decisions? Social Psychological and Personality Science, 8, 493–499. doi:10.1177/1948550617714584

21.

Cuddy

A. J. C.

Schultz

S. J.

Fosse

N. E.

(2018). P-curving a more comprehensive body of research on postural feedback reveals clear evidential value for power-posing effects: Reply to Simmons and Simonsohn (2017). Psychological Science, 29, 656–666. doi:10.1177/0956797617746749

22.

*Cuddy

A. J. C.

Wilmuth

C. A.

Yap

A. J.

Carney

D. R.

(2015). Preparatory power posing affects nonverbal presence and job interview performance. Journal of Applied Psychology, 100, 1286–1295. doi:10.1037/a0038543

23.

Damasio

A. R.

(1999). The feeling of what happens: Body and emotion in the making of conscious-ness. Fort Worth, TX: Harcourt College Publishers.

24.

Damasio

A. R.

Everitt

B. J.

Bishop

(1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 351, 1413–1420. doi:10.1098/rstb.1996.0125

25.

Deeks

J. J.

Macaskill

Irwig

(2005). The performance of tests of publication bias and other sample size effects in systematic reviews of diagnostic test accuracy was assessed. Journal of Clinical Epidemiology, 58, 882–893. doi:10.1016/j.jclinepi.2005.01.016

26.

*de Zavala

A. G.

Lantos

Bowden

(2017). Yoga poses increase subjective energy and state self-esteem in comparison to “power poses.” Frontiers in Psychology, 8, Article 752. doi:10.3389/fpsyg.2017.00752

27.

Duval

Tweedie

(2000). Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56, 455–463.

28.

Eder

A. B.

Rothermund

(2008). When do motor behaviors (mis)match affective stimuli? An evaluative coding view of approach and avoidance reactions. Journal of Experimental Psychology: General, 137, 262–281. doi:10.1037/0096-3445.137.2.262

29.

Egger

Smith

G. D.

Schneider

Minder

(1997). Bias in meta-analysis detected by a simple, graphical test. BMJ, 315, 629–634. doi:10.1136/bmj.315.7109.629

30.

Ekman

(1992). An argument for basic emotions. Cognition and Emotion, 6, 169–200.

31.

Ekman

(2003). Emotions revealed. New York, NY: Time Books.

32.

Farabaugh

Alpert

Wisniewski

S. R.

Otto

M. W.

Fava

Baer

. . . Thase

M. E.

(2012). Cognitive therapy for anxious depression in STARD: What have we learned? Journal of Affective Disorders, 142, 213–218. doi:10.1016/j.jad.2012.04.029

33.

*Fischer

Fischer

Englich

Aydin

Frey

(2011). Empower my decisions: The effects of power gestures on confirmatory information processing. Journal of Experimental Social Psychology, 47, 1146–1154. doi:10.1016/j.jesp.2011.06.008

34.

*Flack

W. F.

(2006). Peripheral feedback effects of facial expressions, bodily postures, and vocal expressions on emotional feelings. Cognition and Emotion, 20, 177–195. doi:10.1080/02699930500359617

35.

*Förster

Stepper

(2000). Compatibility between approach/ avoidance stimulation and valenced information determines residual attention during the process of encoding. European Journal of Social Psychology, 30, 853–871.

36.

Frye

C. A.

Rhodes

M. E.

(2008). The role and mechanisms of steroid hormones in approach-avoidance behavior. In Elliot

A. J.

(Ed.), Handbook of approach and avoidance motivation (pp. 109–126). New York, NY: Psychology Press.

37.

*Garrison

K. E.

Tang

Schmeichel

B. J.

(2016). Embodying power: A preregistered replication and extension of the power pose effect. Social Psychological and Personality Science, 7, 623–630. doi:10.1177/1948550616652209

38.

Gronau

Q. F.

Van Erp

Heck

D. W.

Cesario

Jonas

K. J.

Wagenmakers

E. J.

(2017). A Bayesian model-averaged meta-analysis of the power pose effect with informed and default priors: The case of felt power. Comprehensive Results in Social Psychology, 2, 123–138. doi:10.1080/23743603.2017.1326760

39.

*Hackford

Mackey

Broadbent

(2019). The effects of walking posture on affective and physiological states during stress. Journal of Behavior Therapy and Experimental Psychiatry, 62, 80–87. doi:10.1016/j.jbtep.2018.09.004

40.

*Hao

Xue

Yuan

Wang

Runco

M. A.

(2017). Enhancing creativity: Proper body posture meets proper emotion. Acta Psychologica, 173, 32–40. doi:10.1016/j.actpsy.2016.12.005

41.

*Harmon-Jones

Peterson

C. K.

(2009). Supine body position reduces neural response to anger evocation: Short report. Psychological Science, 20, 1209–1210. doi:10.1111/j.1467-9280.2009.02416.x

42.

Hedges

L. V.

(1981). Distribution theory for Glass’s estimator of effect size and related estimators. Journal of Educational Statistics, 6, 107–128.

43.

Higgins

J. P. T.

Thompson

S. G.

Deeks

J. J.

Altman

D. G.

(2002). Measuring inconsistency in meta-analyses Testing for heterogeneity. BMJ, 327, 557–560.

44.

Hofmann

S. G.

Asnaani

Vonk

I. J. J.

Sawyer

A. T.

Fang

(2012). The efficacy of cognitive behavioral therapy: A review of meta-analyses. Cognitive Therapy and Research, 36, 427–440. doi:10.1007/s10608-012-9476-1

45.

*Huang

Galinsky

A. D.

Gruenfeld

D. H.

Guillory

L. E.

(2011). Powerful postures versus powerful roles: Which is the proximate correlate of thought and behavior? Psychological Science, 22, 95–102. doi:10.1177/0956797610391912

46.

*Jackson

Nault

Richman

L. S.

LaBelle

Rohleder

(2017). Does that pose become you? Testing the effect of body postures on self-concept. Comprehensive Results in Social Psychology, 2, 81–105.

47.

James

(1994). The physical basis of emotion. Psychological Review, 101, 205–210.

48.

Jonas

K. J.

Cesario

Alger

Bailey

A. H.

Bombari

Carney

. . . Tybur

J. M.

(2017). Power poses—Where do we stand? Comprehensive Results in Social Psychology, 2, 139–141. doi:10.1080/23743603.2017.1342447

49.

Josephs

R. A.

Sellers

J. G.

Newman

M. L.

Mehta

P. H.

(2006). The mismatch effect: When testosterone and status are at odds. Journal of Personality and Social Psychology, 90, 999–1013. doi:10.1037/0022-3514.90.6.999

50.

Kabat-Zinn

(2005). Coming to our senses: Healing ourselves and the world through mindfulness. New York, NY: Hyperion.

51.

*Keller

V. N.

Johnson

D. J.

Harder

J. A.

(2017). Meeting your inner super(wo)man: Are power poses effective when taught? Comprehensive Results in Social Psychology, 53, 1689–1699. doi:10.1017/CBO9781107415324.004

52.

*Kille

D. R.

Forest

A. L.

Wood

J. V.

(2012). Tall, dark, and stable: Embodiment motivates mate selection preferences psychological. Psychological Science, 24, 112–114. doi:10.1177/0956797612457392

53.

*Klaschinski

Schnabel

Schröder-Abé

(2017). Benefits of power posing: Effects on dominance and social sensitivity. Comprehensive Results in Social Psychology, 2, 55–67. doi:10.1080/23743603.2016.1248080

54.

*Kozak

M. N.

Roberts

T. A.

Patterson

K. E.

(2014). She stoops to conquer? How posture interacts with self-objectification and status to impact women’s affect and performance. Psychology of Women Quarterly, 38, 414–424. doi:10.1177/0361684313517865

55.

Lakoff

G. J.

Johnson

(1999). Philosophy in the flesh: The embodied mind and its challenge to Western thought. New York, NY: Basic Books.

56.

*Latu

I. M.

Duffy

Pardal

Alger

(2017). Power vs. persuasion: Can open body postures embody openness to persuasion? Comprehensive Results in Social Psychology, 2, 68–80. doi:10.1080/23743603.2017.1327178

57.

*Lee

E. H.

Schnall

(2014). The influence of social power on weight perception. Journal of Experimental Psychology: General, 143, 1719–1725. doi:10.1037/a0035699

58.

S. T.

Abelson

J. L.

Okada

Taylor

S. F.

Liberzon

(2017). Neural circuitry of emotion regulation: Effects of appraisal, attention, and cortisol administration. Cognitive, Affective and Behavioral Neuroscience, 17, 437–451. doi:10.3758/s13415-016-0489-1

59.

Maiese

(2014). How can emotions be both cognitive and bodily? Phenomenology and the Cognitive Sciences, 13, 513–531. doi:10.1007/s11097-014-9373-z

60.

Merleau-Ponty

(1962). Phenomenology of perception. London, England: Routledge & Kegan Paul.

61.

Michalak

Burg

Heidenreich

(2012). Don’t forget your body: Mindfulness, embodiment, and the treatment of depression. Mindfulness, 3, 190–199. doi:10.1007/s12671-012-0107-4

62.

Michalak

Chatinyan

Chourib

Teismann

(2018). The impact of upward versus downward movement patterns on memory characteristics of depressed individuals. Psychopathology, 51, 326–334. doi:10.1159/000492788

63.

Michalak

Mischnat

Teismann

(2014). Sitting posture makes a difference-embodiment effects on depressive memory bias. Clinical Psychology and Psychotherapy, 21, 519–524. doi:10.1002/cpp.1890

64.

*Michalak

Rohde

Troje

N. F.

(2015). How we walk affects what we remember: Gait modifications through biofeedback change negative affective memory bias. Journal of Behavior Therapy and Experimental Psychiatry, 46, 121–125. doi:10.1016/j.jbtep.2014.09.004

65.

*Mussweiler

(2006). Doing is for thinking! Stereotype activation by stereotypic movements. Psychological Science, 17, 17–21. doi:10.1111/j.1467-9280.2005.01659.x

66.

*Nair

Sagar

Sollers

Consedine

Broadbent

(2015). Do slumped and upright postures affect stress responses? A randomized trial. Health Psychology, 34, 632–641. doi:10.1037/hea0000146

67.

*Naruse

Hirai

(2000). Effects of slow tempo exercise on respiration heart rate and mood state. Perceptual and Motor Skills, 91, 729–740.

68.

Newman

M. G.

Castonguay

L. G.

Borkovec

T. D.

Fisher

A. J.

Boswell

J. F.

Szkodny

L. E.

Nordberg

S. S.

(2011). A randomized controlled trial of cognitive-behavioral therapy for generalized anxiety disorder with integrated techniques from emotion-focused and interpersonal therapies. Journal of Consulting and Clinical Psychology, 79, 171–181. doi:10.1037/a0022489

69.

Niedenthal

P. M.

(2007). Embodying emotion. Science, 316, 1002–1005. doi:10.1126/science.1136930

70.

Niedenthal

P. M.

Brauer

(2012). Social functionality of human emotion. doi:10.1146/annurev.psych.121208.131605

71.

*Park

L. E.

Streamer

Huang

Galinsky

A. D.

(2013). Stand tall, but don’t put your feet up: Universal and culturally-specific effects of expansive postures on power. Journal of Experimental Social Psychology, 49, 965–971. doi:10.1016/j.jesp.2013.06.001

72.

*Peper

Lin

I.-M.

(2012). Increase or decrease depression: How body postures influence your energy level. Biofeedback, 40, 125–130. doi:10.5298/1081-5937-40.3.01

73.

*Peper

Lin

I.-M.

Harvey

Perez

(2017). How posture affects memory recall and mood. Biofeedback, 45, 36–41. doi:10.5298/1081-5937-45.2.01

74.

Poole

Greenland

(1999). Random-effects meta-analyses are not always conservative. American Journal of Epidemiology, 150, 469–475. doi:10.1093/oxfordjournals.aje.a01003

75.

*Ranehill

Dreber

Johannesson

Leiberg

Sul

Weber

R. A.

(2015). Assessing the robustness of power posing: No effect on hormones and risk tolerance in a large sample of men and women. Psychological Science, 26, 653–656. doi:10.1177/0956797614553946

76.

*Riskind

J. H.

(1984). They stoop to conquer: Guiding and self-regulatory functions of physical posture after success and failure. Journal of Personality and Social Psychology, 47, 479–493. doi:10.1037/0022-3514.47.3.479

77.

*Riskind

J. H.

Gotay

C. C.

(1982). Physical posture: Could it have regulatory or feedback effects on motivation and emotion? Motivation and Emotion, 6, 273–298.

78.

Roald

Levin

Køppe

(2018). Affective incarnations: Maurice Merleau-Ponty’s challenge to bodily theories of emotion. Journal of Theoretical and Philosophical Psychology, 38, 205–218. doi:10.1037/teo0000101

79.

*Roberts

T. A.

Arefi-Afshar

(2007). Not all who stand tall are proud: Gender differences in the proprioceptive effects of upright posture. Cognition and Emotion, 21, 714–727. doi:10.1080/02699930600826432

80.

Roelofs

van Peer

Berretty

Jong

de Spinhoven

Elzinga

B. M.

(2009). Hypothalamus-pituitary-adrenal axis hyperresponsiveness is associated with increased social avoidance behavior in social phobia. Biological Psychiatry, 65, 336–343. doi:10.1016/j.biopsych.2008.08.022

81.

*Ronay

Tybur

J. M.

van Huijstee

Morssinkhof

(2017). Embodied power, testosterone, and overconfidence as a causal pathway to risk-taking. Comprehensive Results in Social Psychology, 2, 28–43. doi:10.1080/23743603.2016.1248081

82.

Rosenthal

(1991). Applied social research methods series: Meta-analytic procedures for social research. Thousand Oaks, CA: SAGE. doi:10.4135/9781412984997

83.

*Rossberg-Gempton

Poole

G. D.

(1993). The effect of open and closed postures on pleasant and unpleasant emotions. The Arts in Psychotherapy, 20, 75–82.

84.

Scherer

K. R.

Moors

(2019). The emotion process: Event appraisal and component differentiation. Annual Review of Psychology, 70, 719–745. doi:10.1146/annurev-psych-122216-011854

85.

Schubert

T. W.

Koole

S. L.

(2009). The embodied self: Making a fist enhances men’s power-related self-conceptions. Journal of Experimental Social Psychology, 45, 828–834. doi:10.1016/j.jesp.2009.02.003

86.

*Shafir

Taylor

S. F.

Atkinson

A. P.

Langenecker

S. A.

Zubieta

J. K.

(2013). Emotion regulation through execution, observation, and imagery of emotional movements. Brain and Cognition, 82, 219–227. doi:10.1016/j.bandc.2013.03.001

87.

Shapiro

(2011). Embodied cognition. New York, NY: Routledge.

88.

Simmons

J. P.

Simonsohn

(2017). Power posing: P- curving the evidence. Psychological Science, 28, 687–693. doi:10.1177/0956797616658563

89.

*Smith

K. M.

Apicella

C. L.

(2017). Winners, losers, and posers: The effect of power poses on testosterone and risk-taking following competition. Hormones and Behavior, 92, 172–181. doi:10.1016/j.yhbeh.2016.11.003

90.

*Stepper

Strack

(1993). Proprioceptive determinants of emotional and nonemotional feelings. Journal of Personality and Social Psychology, 64, 211–220.

91.

*Strelan

Weick

Vasiljevic

(2014). Power and revenge. British Journal of Social Psychology, 53, 521–540. doi:10.1111/bjso.12044

92.

*Tiedens

L. Z.

Fragale

A. R.

(2003). Power moves: Complementarity in dominant and submissive nonverbal behavior. Journal of Personality and Social Psychology, 84, 558–568. doi:10.1037/0022-3514.84.3.558

93.

*Turan

(2015). Is a submissive posture adaptive when being evaluated negatively? Effects on cortisol reactivity. Neuroendocrinology Letters, 36, 394–398.

94.

*Veenstra

Schneider

I. K.

Koole

S. L.

(2017). Embodied mood regulation: The impact of body posture on mood recovery, negative thoughts, and mood-congruent recall. Cognition and Emotion, 31, 1361–1376. doi:10.1080/02699931.2016.1225003

95.

*Welker

K. M.

Oberleitner

D. E.

Cain

Carré

J. M.

(2013). Upright and left out: Posture moderates the effects of social exclusion on mood and threats to basic needs. European Journal of Social Psychology, 43, 355–361. doi:10.1002/ejsp.1944

96.

*Wilkes

Kydd

Sagar

Broadbent

(2017). Upright posture improves affect and fatigue in people with depressive symptoms. Journal of Behavior Therapy and Experimental Psychiatry, 54, 143–149. doi:10.1016/j.jbtep.2016.07.015

97.

Wilson

A. D.

Golonka

(2013). Embodied cognition is not what you think it is. Frontiers in Psychology, 4, Article 58. doi:10.3389/fpsyg.2013.00058

98.

Wilson

(2002). Six views of embodied cognition. Psychometric Bulletin & Review, 9, 625–636.

99.

*Wilson

V. E.

Peper

(2004). The effects of upright and slumped postures on the recall of positive and negative thoughts. Applied Psychophysiology Biofeedback, 29, 189–195. doi:10.1023/B:APBI.0000039057.32963.34

100.

Winkielman

Niedenthal

Wielgosz

Eelen

Kavanagh

L. C.

(2014). Embodiment of cognition and emotion. In Mikulincer

Shaver

P. R.

(Ed.), APA handbook of personality and social psychology, Vol. 1: Attitudes and social cognition (pp. 151–175). Washington, DC: American Psychological Association. doi:10.1037/14341-004

101.

*Yap

A. J.

Cuddy

A. J. C.

Carney

D. R.

Lucas

B. J.

Wazlawek

A. S.

(2013). The ergonomics of dishonesty. Psychological Science, 24, 2281–2289. doi:10.1177/0956797613492425

102.

Young

Korszun

(2010). Sex, trauma, stress hormones and depression. Molecular Psychiatry, 15, 23–28. doi:10.1038/mp.2009.94